AI & Energy: Inside the Defining Relationship of the Century

The AI boom is reshaping energy demand in ways the grid has not previously experienced.

While AI holds clear potential to enhance grid forecasting, improve renewable integration and enable real-time balancing, the physical infrastructure supporting it is driving a sharp rise in electricity consumption.

In many cases, this demand is still being met by gas generation, raising concerns about long-term sustainability.

These competing dynamics were explored during the “Energy and AI. Who’s working for who?” fireside chat on Day 2 of Data Centre LIVE: The London Summit.

Led by Dr. Ben Krikler, Head of Energised Futures and Director of Research & Innovation at Centrica, and moderated by Data Centre Magazine's Ben Craske, the session unpacked how energy systems are being reshaped by AI-driven demand.

One of the biggest takeaways from the session was the fact that AI is both intensifying pressure on the grid and offering tools to help manage it, creating a push-pull tension.

“It’s clearly a hungry beast,” Ben said, pointing to projections that AI-related energy demand could quadruple over the next decade.

"At the same time, AI could become one of the energy sector’s most powerful optimisation tools, improving forecasting, supporting decarbonisation efforts and helping operators make smarter infrastructure decisions in real time."

Rethinking data centre power with microgrids

A key energy-focused innovation discussed was Centrica’s work on DC-connected microgrids.

Ben highlighted their potential to significantly improve how power is distributed and used within data centres, revisiting the longstanding DC versus AC debate through an efficiency lens.

Much of today’s data centre equipment, including batteries, solar systems and compute hardware, already operates on DC power. Repeated conversion between AC and DC introduces avoidable energy losses.

By reducing these conversions, efficiency gains of 15–20% could be achieved.

“It’s a sizable benefit,” he explained.

Flexibility also emerged as a central theme in managing energy demand.

Operators are attempting to forecast AI-driven consumption years in advance, often building capacity for workloads that have yet to materialise. In this context, flexibility is becoming a core energy strategy.

The ability to shift workloads geographically, delay non-urgent AI training or reduce consumption during peak grid stress could help operators minimise connection requirements and speed up deployment.

In reality, flexibility could become one of the defining tools of future grid management.

Importantly, Ben noted that these adjustments need not compromise reliability.

“There’s lots of evidence,” he noted, “that there’s no impact on critical loads.”

Aligning energy systems with AI growth

Perhaps the most striking insight came through Ben's analogy describing the disconnect between AI innovation and energy infrastructure development.

AI, he said, behaves like a fast-moving river, constantly evolving and difficult to predict.

Energy infrastructure, by contrast, resembles a glacier, progressing slowly and requiring long-term certainty and capital investment.

Regulation, meanwhile, operates like tectonic plates.

“They don’t move at the speed you’d like,” he said.

This misalignment is becoming increasingly visible in grid connection bottlenecks. Demand from data centres has surged in recent months, with many operators exploring on-site energy generation to bypass delays.

The sector now faces the challenge of building energy-intensive infrastructure for future applications that remain uncertain, while under pressure to connect capacity faster than ever.

Managing perception

Beyond technology, the discussion also turned to how rising energy demand is perceived by the public.

Centrica’s Energised Futures team combines energy research with behavioural insights to understand how communities respond to infrastructure expansion.

For data centre operators, public perception is becoming a critical risk, particularly where increased demand is associated with fossil fuel generation or local grid strain.

“There’s a very big risk that we damage the perception of this whole industry,” he warned.

Rather than relying on large-scale campaigns, Ben suggested that straightforward engagement could make a meaningful difference. Collaborating with local businesses and communities, as well as schools, can help demystify data centres and their role in modern energy systems.

He also cautioned against fully “islanded” developments that operate in isolation from surrounding communities, as these could deepen mistrust.

Instead, he advocated for integration through initiatives such as local heat reuse, shared energy systems and community partnerships.

AI’s role in future energy systems

Despite the scale of the challenge, the session closed with optimism about AI’s role in shaping future energy systems.

Ben pointed to advances in battery chemistry, fuel cells and storage optimisation as areas where AI could unlock significant progress.

“If AI can help us with storage,” he said, “that could have huge potential.”

Looking ahead, the concept of “energy communities” is gaining traction. These decentralised systems would allow homes, electric vehicles, businesses and data centres to share energy locally, rather than relying solely on centralised grids.

In this model, AI becomes more than just a major energy consumer. It becomes an active participant in balancing supply and demand.

Whether energy infrastructure can evolve quickly enough to support this shift remains an open question, and one that will define the sector’s trajectory in the years ahead.